Bin
Wei
a,
Justin C.
Tokash
a,
Guang
Chen
ab,
Michael A.
Hickner
b and
Bruce E.
Logan
*a
aDepartment of Civil and Environmental Engineering, Penn State University, University Park, PA 16802, USA. E-mail: blogan@psu.edu; Fax: +1-814-863-7304
bDepartment of Materials Science and Engineering, Penn State University, University Park, PA 16802, USA
First published on 4th October 2012
Activated carbon (AC) air cathodes were constructed using variable amounts of carbon (43–171 mg cm−2) and an inexpensive binder (10 wt% polytetrafluoroethylene, PTFE), and with or without a porous cloth wipe-based diffusion layer (DL) that was sealed with PDMS. The cathodes with the highest AC loading of 171 mg cm−2, and no diffusion layer, produced 1255 ± 75 mW m−2 and did not appreciably vary in performance after 1.5 months of operation. Slightly higher power densities were initially obtained using 100 mg cm−2 of AC (1310 ± 70 mW m−2) and a PDMS/wipe diffusion layer, although the performance of this cathode decreased to 1050 ± 70 mW m−2 after 1.5 months, and 1010 ± 190 mW m−2 after 5 months. AC loadings of 43 mg cm−2 and 100 mg cm−2 did not appreciably affect performance (with diffusion layers). MFCs with the Pt catalyst and Nafion binder initially produced 1295 ± 13 mW m−2, but the performance decreased to 930 ± 50 mW m−2 after 1.5 months, and then to 890 ± 20 mW m−2 after 5 months. Cathode performance was optimized for all cathodes by using the least amount of PTFE binder (10%, in tests using up to 40%). These results provide a method to construct cathodes for MFCs that use only inexpensive AC and a PTFE, while producing power densities similar to those of Pt/C cathodes. The methods used here to make these cathodes will enable further tests on carbon materials in order to optimize and extend the lifetime of AC cathodes in MFCs.
Catalysts can be used to reduce the activation energy needed for oxygen reduction at the cathode. Noble metals such as platinum are most commonly used, but these catalysts are very expensive (e.g. $140/g for Pt). Other metal-based catalysts (cobalt-tetramethylphenylporphyrin, pyrolyzed-Fe(II) phthalocyanine, and manganese oxides) have been used in MFC cathodes to achieve performance similar to that of platinum.7–10 Cheng et al.7 determined that platinum loadings of 0.1–0.5 mg cm−2 did not increase MFC performance with the addition of more platinum. The loading of non-noble metal based catalysts, however, are much higher than those needed for Pt, which can significantly increase the costs for making these cathodes with other metal catalysts.
Activated carbon (AC) is a promising alternative to Pt for MFC applications. AC is very inexpensive ($2.6/kg) and cathodes made with AC have achieved nearly the same performance as those made using Pt at current densities relevant for MFC applications.11 For example, at an applied potential of −200 mV (versus Ag/AgCl), Pant et al.12 obtained −0.6 mA cm−2 for a non-platinized AC electrode, compared to −0.75 mA cm−2 for Pt on graphite, and −0.43 mA cm−2 for a non-platinized graphite control. AC cathodes have been made using a polytetrafluoroethylene (PTFE) binder and a 70% porous PTFE diffusion layer on nickel mesh current collector by a proprietary process.13 MFCs using these cathodes produced a maximum power density of 1220 mW m−2 in MFC tests, compared to 1060 mW m−2 using Pt on carbon cloth.11 More recently a rolling method with sintering at 340 °C was used to make AC cathodes, with PTFE used as both the binder and the diffusion layer.14 Power densities were lower than those previously reported11,15 for AC cathodes, ranging from 802 to 584 mW m−2, depending on the AC/PTFE ratio. AC loading was not examined.
A major limitation of both Pt and AC cathodes is that their performance decrease over time. Zhang et al.15 found that maximum power densities of AC cathodes decreased from 1214 ± 123 mW m−2 to 750 mW m−2 after one year of MFC operation. Similar decreases in performance have been shown for AC cathodes with a less porous PTFE diffusion layer (30%), and for Pt-catalyzed cathodes using various types of binders.16 One approach to improve longevity of the AC cathodes could be to increase the AC loading, as AC is inexpensive and this would not appreciably affect overall costs. However, previously examined AC cathodes were made using a proprietary pressing process, and therefore the methods to construct them with various AC loadings are not known.12,13,15 Other AC cathodes using a rolling and sintering process had lower performance, and were not compared against more traditional Pt catalyst cathodes.17
The objectives of this study were to develop a method to make AC cathodes to enable their routine analysis in the laboratory, and to examine the effects of AC loading and binder content on cathode performance over time. An increase in AC loading was expected to increase surface area and active sites for oxygen reduction, which could lead to lower cathodic overpotentials, improved power densities, and extended performance. Although Nafion is often used as a cathode binder, PTFE is less expensive than Nafion, and cathodes made with PTFE have produced similar performance to those made with Nafion and a Pt catalyst.18,19 The performance of the AC cathodes made here were benchmarked against cathodes made using more typical but much more expensive materials of carbon cloth, Pt, Nafion, and PTFE diffusion (four applications) layers.
AC cathodes were prepared with AC powder (Norit SX plus, Norit Americas Inc., TX) and a PTFE binder (in 60% emulsion), at a weight ratio of AC to PTFE of 9:1 except as noted. After the AC and PTFE solution mixture formed a paste, it was spread onto one side of a stainless steel mesh current collector (50 × 50, type 304, McMaster-Carr, OH) using a spatula, and then pressed at 10,000 pounds-force for 20 min (Carver press, Model 4386, Carver Inc., IN, USA), all at room temperature. AC cathodes were examined for power production with and without diffusion layers. A poly(dimethylsiloxane) (PDMS) solution was prepared as described.21 Diffusion layers were prepared by applying 2 layers of PDMS solution onto a cloth wipe material (DuPont Sontara, Style 8864), as opposed to application directly onto the cathode as previously tested,21 in order to minimize clogging of the cathode catalyst with the PDMS. AC cathodes were dried at 70–90 °C overnight in an oven before use. AC cathodes with diffusion layers were made with the following loadings (projected area of 7 cm2): 7, 11, 14, 28, 43, 100, 171 mg cm−2 with 10% PTFE binder. Higher AC loadings of 43, 71, 100, 143, 171 mg cm−2 were used in tests with cathodes lacking a diffusion layer. Cathodes without diffusion layers were also tested that had different PTFE binder contents of 10, 15, 20, 25, 30, 35 and 40 wt%, with 43 mg cm−2 of AC.
The AC cathodes were compared to carbon cloth cathodes (CC, 30% wet proofing, Fuel Cell Earth LLC) containing a platinum catalyst (0.5 mg cm−2) held with carbon black (Vulcan XC-72) and a Nafion binder on the solution side as a control.22 This cathode had PTFE diffusion layers that were applied on the air side, to be consistent with previous studies.23
Cathode current–voltage (polarization) curves were examined by applying cathode potentials in a stepwise manner after operating under open circuit conditions for 3 h. Each potential (0.41 V, 0.31 V, 0.21 V, 0.11 V and 0.01 V, versus SHE) was applied for 2 h, in order to ensure a steady state current.
Polarization curves were obtained using different resistors, with one external resistor used for a complete fed batch cycle (multi-cycle method). Polarization data were obtained after the MFCs had been operated at 1000 Ω for 1.5 and 5 months, by varying the external resistances from 1000 Ω to 50 Ω. Current densities were normalized to the cathode projected area (7 cm2) and calculated using i = E/RextA where E was the voltage measured when MFCs reached a steady-state condition. Data was recorded using a data acquisition system (2700, Keithley instruments, OH). Power densities were calculated with P = iE/A. Coulombic efficiencies were measured at each external resistance as previously described.1 All MFC tests were carried out in duplicate.
Electrochemical impedance spectroscopy (EIS) was conducted at a set potential of either 0.3 V or 0.1 V, over a frequency range of 0.5 MHz to 1 mHz, with a sinusoidal perturbation of 10 mV. EIS results were analyzed by fitting the spectra to an equivalent circuit model (†. S1, ESI†). Since there was a high frequency semicircle in the Nyquist plots for AC cathodes which was not observed in those for Pt/C cathodes (†. S2, ESI†), two equivalent circuits in series were fit to Nyquist plots obtained with AC cathodes and Pt/C cathodes using the potentiostat software (EC-lab). The circuits contained resistors for the solution resistance (Rs), contact resistance (Rc), charge transfer resistance (Rct), in addition to an Warburg impedance (Zw) as the diffusive element when the diffusion layer thickness on the solution-electrode interface became semi-infinite24 and constant phase elements (Q1 and Q2) which describe non-ideal capacitive properties since the a factor was not equal to 1 (for ideal capacitance a factor is 1). The definition of Warburg impedance is:25
Zw = σw(ω)−1/2(1 − j) | (1) |
The Warburg coefficient σw is:26
(2) |
In the above equations, ω is the radial frequency, DO and DR the diffusion coefficients of the oxidant and reductant, A the surface area of the electrode, n the numbers of electrons evolved in the redox reaction, σw the Warburg coefficient, CO and CR the bulk concentrations of oxidant and reductant, F Faraday's constant, R the gas constant, and T the absolute temperature. The total cathode polarization resistance was obtained from the first derivative of the voltage-current (polarization) curves using a second order polynomial function regression.27 The diffusion resistance was calculated as the difference between this total resistance and the solution, contact and charge transfer resistances.
Galvanostatic charge-discharge tests were carried out with clean cathodes at a current density of 0.35 mA cm−2 by discharging the cathode for 2 h, followed by charging for 2 h, and then discharging for 3 h. Cathode specific capacitance with activated carbon cathode was normalized to the weight of catalyst and calculated from Cm = IΔt/ΔVm, with the slope (ΔV/Δt) of the linear region below OCV (open circuit voltage, ca. 0.46 V for AC cathodes).
Fig. 1 Current–voltage (polarization) curves using (a) AC cathodes without diffusion layers (43, 71, 100, 142, 171 mg cm−2 AC, 10% PTFE) compared to Pt/C cathodes; (b) activated carbon cathodes with diffusion layers (7, 11, 14, 28, 43, 100, 171 mg cm−2 AC, 10% PTFE), compared to Pt/C cathodes; (c) activated carbon cathodes with different PTFE contents (10, 15, 25, 35, 40 wt %, with 43 mg cm−2 AC). |
AC cathodes made with different amounts of binder (10–40% PTFE) did not show any consistent trend between binder content and current at cathode potentials ranging from 0 V to the OCV (ca. 0.46 V) (Fig. 1). This result suggested PTFE binder content was not an important factor in AC cathode construction, and that the use of binder could be minimized to reduce overall cathode costs.
Fig. 2 Maximum power densities obtained using (a) AC cathodes (7, 11, 14, 28, 42, 100, 171 mg cm−2 AC, 10% PTFE) with diffusion layer, and Pt/C cathodes in the initial cycles, and after 1.5 and 5 months operation (nt: not tested); (b) AC cathodes (43, 71, 100, 143, 171 mg cm−2 AC, 10% PTFE) with diffusion layer, and Pt/C cathodes in the initial cycles; (c) AC cathodes (20, 25, 30, 40% PTFE, 43 mg cm−2 AC) without diffusion layer, and Pt/C cathodes after 1 month. |
Fig. 3 Cathode and anode polarization curves obtained using (a) AC cathodes (43, 71, 100, 143, 171 mg cm−2 AC, 10% PTFE) without DL, Pt/C cathodes in the initial cycles; (b) AC cathodes (7, 11, 14, 28, 43, 100, 171 mg cm−2 AC, 10% PTFE) with DL, Pt/C cathodes in the initial cycles; (c) AC cathodes (7, 11, 28, 43, 100, 171 mg cm−2 AC, 10% PTFE) with DL, Pt/C cathodes after 1.5 months; (d) AC cathodes (11, 28, 43, 100, 171 mg cm−2 AC, 10% PTFE) after 5 months. |
When AC cathodes were made without diffusion layers, the cathodes with the highest AC carbon loading of 171 mg cm−2 produced the highest power density of 1255 ± 75 mW m−2, which was similar to that obtained with the Pt/C cathode (1300 ± 15 mW m−2) (Fig. 2). The other AC cathodes produced slightly lower power densities. Anodic polarization curves were similar, indicating different power densities were due to cathode performance.
AC cathodes with different PTFE binder contents (20–40%) had similar maximum power densities to each other (ca. 1100 mW m−2) and to the Pt/C cathodes (985 ± 70 mW m−2) after 1 month (Fig. 2). The anodes performed quite comparably and there was no significant difference in cathode performance with the variable amount of PTFE binder (Fig. 4), consistent with abiotic electrochemical test results (Fig. 1).
Fig. 4 Cathode and anode polarization curves obtained using AC cathodes (20, 25, 30, 35, 40% PTFE, 43 mg cm−2 AC) without DL, Pt/C cathodes after 1 month. |
An increase in the AC catalyst loading from 7 to 100 mg cm−2 generally led to improved current generation at cathode potentials relevant for MFC operation. In one case, however, the cathode containing the highest amount of AC (171 mg cm−2), when it had a diffusion layer, had a lower current than cathodes with 100 mg cm−2 in abiotic electrochemical tests (Fig. 1). New cathodes with 100 mg cm−2 AC initially produced the highest maximum power density of 1310 ± 70 mW m−2, similar to that obtained with the Pt/C cathode (1300 ± 15 mW m−2) (Fig. 2). In general, the maximum power density increased with AC loading. However, an increase in the loading from 100 to 171 mg cm−2 did not improve performance, perhaps as a result of larger distance over which proton and oxygen gradients can form across the cathode as the catalyst layer became thicker.
Fig. 5 Coulombic efficiencies (CEs) obtained using (a) AC cathodes (7, 11, 14, 28, 43, 100, 171 mg cm−2 AC, 10% PTFE) with DL, Pt/C cathodes in the initial cycles; (b) AC cathodes (7, 11, 28, 43, 100, 171 mg cm−2 AC, 10% PTFE) with DL, Pt/C cathodes after 1.5 months; (c) AC cathodes (11, 28, 43, 100, 171 mg cm−2 AC, 10% PTFE) with DL, Pt/C cathodes after 5 months; (d) AC cathodes (43, 71, 100, 143, 171 mg cm−2 AC, 10% PTFE) without DL, Pt/C cathodes in the initial cycles. |
Fig. 6 Resistance components obtained at cathode potential of (a) 0.3 V; (b) 0.1 V using AC cathodes (11, 14, 43, 100, 171 mg cm−2 AC, 10% PTFE) with DL, and Pt/C cathodes; (c) 0.1 V using AC cathodes (10, 15, 20, 25, 30, 35, 40% PTFE, 43 mg cm−2 AC) without DL. |
The largest part of the resistance was the diffusion resistance (Fig. 6). This component contributed between 45 and 75% of the total internal resistance at 0.3 V, and 47 to 63% at 0.1 V. At 0.3 V, the charge transfer resistances were similar for cathodes with 11–28 mg cm−2 AC (12.7, 14.9 and 13.1 Ω), and 77% lower for higher AC amounts (43 mg cm−2, 3.0 Ω and 100 mg cm−2, 3.0 Ω). However, the Rct increased to 10.5 Ω with the highest AC loading (171 mg cm−2). These results showed carbon loadings of 43 and 100 mg cm−2 had the higher reaction rates.
The Warburg impedance decreased inversely with AC loading from 11 to 100 mg cm−2 (Zw of 2.6–0.2 Ω s−1/2), indicating faster mass transport processes (larger diffusivities of redox species) to the catalyst. At 0.1 V, Rct slightly increased and Zw decreased with AC cathodes compared to measurements at 0.3 V. When the cathode potential was 0.1 V, cathodes with lower AC loadings (11–28 mg cm−2 AC) had higher charge transfer and diffusion resistances than cathodes with more AC (43–171 mg cm−2), consistent with the results obtained at 0.3 V. Moreover, when the cathode potential was reduced from 0.3 V to 0.1 V, charge transfer decreased as a result of higher driving force for the cathode reactions, but diffusion resistance increased. Compared to AC cathodes, Pt/C cathodes had lower Rct and Zw values at 0.1 V compared to 0.3 V. The change in Rct at the two different cathode potentials was consistent with a previous study28 that employed a 1 M H2SO4 electrolyte and carbon electrodes prepared with porous carbon powders and PVDF binder. The increase in charge transfer resistance could lead to the reduction in the Warburg impedance (Zw) with a slower reaction rate, as this situation requires less reactant for the oxygen reduction reaction.
Analysis of the cathodes with different percentages of PTFE binder (at 0.1 V) using EIS showed that the overall resistances were generally similar (Fig. 6). The cathodes with the two highest binder contents (35 and 40%) had slightly higher total resistances, and much larger charge transfer resistances than the cathodes with less PTFE. These data suggest that a higher binder content reduced contact between the current collector and the AC particles. However, the analysis was inconclusive because the diffusion resistance decreased appreciably for these two samples (Table S2, ESI†), and therefore the contributions of the diffusion and charge transfer resistances were not well separated in this circuit analysis.
Polymeric binders are used to hold a catalyst to the conductive material, while at the same time the ionic nature a binder, such as Nafion, is designed to enable ion transport from the bulk solution to catalytic reaction sites. Hydrophobic binders have been demonstrated in MFCs with good results,7 however, too much hydrophobic binder can reduce the porosity of the cathode and inhibit ion transport to the electrochemically active surface area.29 In cathodes with different PTFE binder contents tested here, Rct increased with binder content, likely as a result of an excess of PTFE reducing the available surface area for oxygen reduction. The specific capacitance of the AC, however, showed only slight changes with AC loading with a constant amount of binder. This result suggests that the capacitance of the AC was not an important factor for cathode design. Instead, it is more critical to minimize the amount of binder in order to reduce charge transfer resistance.
Fig. 7 Charge-discharge diagram using AC cathodes (43, 71, 100, 143, 171 mg cm−2 AC, 10% PTFE) and Pt/C cathodes. |
Footnote |
† Electronic Supplementary Information (ESI) available. See DOI: 10.1039/c2ra21572a |
This journal is © The Royal Society of Chemistry 2012 |